Google publicly released a massive new set of data, collected from users’ mobile devices, showing where stay-at-home directives during the coronavirus pandemic are working — and where they’re not.

On Friday, Google published what it called an early release of its COVID-19 Community Mobility Reports (available at google.com/covid19/mobility). The data spans 131 countries and regions, including a breakdown for all 50 U.S. states, Washington, D.C., and Puerto Rico.

Google said it is making the anonymized user data available in an effort to help public health officials gain visibility into the COVID-19 pandemic and determine the efficacy of social-distancing measures to slow the spread of the virus.

The data tracks user visits over time in six high-level categories: retail and recreation (including movie theaters, theme parks, restaurants and shopping centers); grocery and pharmacy; parks; transit stations; workplaces; and residential.

For the U.S. overall from Feb. 16-March 29, the Google data shows a 47% drop in visits to retail and recreation sites, a 38% decline in workplace, and a 12% increase in residential. There are significant regional differences: In New York, which has become the epicenter of the pandemic in the U.S., user data for Manhattan over the same time period shows an 86% plunge for retail and recreation, -57% for workplace and +19% for residential. In Los Angeles County, retail and recreation visits are -51%; workplace is -41%; and residential is +17%.

Google emphasized that it is not “at any point” disclosing any personally identifiable information (such as an individual’s location, contacts or movement). The data it is reporting uses a differential privacy methodology based on information gathered from Google users who have turned on the “Location History” setting, which is disabled by default.

The internet giant’s COVID-19 reports use aggregated, anonymized data to chart movement trends over time by geography, showing trends over several weeks with the most recent information reflecting the last 48-72 hours. For some local geographies, Google does not have sufficient data to provide a meaningful analysis.

In a blog post, Google execs expressed hope the information could help officials improve recommendations for business hours and delivery services, as well as predict demand for public transportation and optimize for social distancing. Google said it also is working with “select epidemiologists” to better forecast the COVID-19 pandemic.

Facebook last year launched a similar project, Disease Prevention Maps, which is available to nonprofit organizations and researchers. Announced in May 2019, it uses anonymized Facebook user data to track the movement and connectivity of populations in geographic areas.